We previously used B/HPIV3 to express a prefusion-stabilized version of the full-length SARS-CoV-2 S protein (Wuhan-Hu-1, GenBank MN908947), codon-optimized for human expression, and with the S1/S2 polybasic furin cleavage motif “RRAR” ablated by aa substitutions (RRAR-to-GSAS; [23]), resulting in B/HPIV3/S-2P [17]. Here, we generated an additional version, B/HPIV3/S-6P, expressing a full-length version of the SARS-CoV-2 S protein that had been further stabilized by four additional proline substitutions (aa 817, 892, 899 and 942; [21], Fig 1A). For expression in B/HPIV3, the open reading frames (ORFs) encoding the prefusion-stabilized S proteins were framed by nucleotide adapters containing the BPIV3 gene start and gene end signals for insertion as an additional gene between the BPIV3 N and P ORFs (Fig 1A).

B/HPIV3 expressing the S-6P prefusion-stabilized version of the S protein was readily recovered from cDNA by reverse genetics. The sequences of all recombinant B/HPIV3 viruses used in this study were confirmed by Sanger sequencing. We next compared the stability of S expression of S-2P and S-6P by dual-staining immunoplaque assay, performed with polyclonal hyperimmune antisera against PIV3 antigens (red pseudo-coloring), and against the recombinantly-expressed secreted version of the SARS-CoV-2 S-2P protein (green pseudo-coloring). B/HPIV3/S-2P plaques appeared heterogeneous in size, while B/HPIV3/S-6P plaques were more homogenous (Fig 1B). In virus stocks grown from eight independent recoveries of B/HPIV3/S-2P or B/HPIV3/S-6P, staining for both PIV3 and SARS-CoV-2 S (yellow, Fig 1B) was obtained for 94.9±3.4% and 96.7±1.7% of plaques, respectively, indicating that both viruses stably expressed SARS-CoV-2 S protein. In Vero cells, which are widely used for vaccine manufacture, multicycle replication of B/HPIV3/S-2P and B/HPIV3/S-6P was similarly efficient and only slightly delayed compared to B/HPIV3 (Fig 1C), while in human lung epithelial A549 cells, replication of both B/HPIV3/S-2P and B/HPIV3/S-6P was reduced by about 10-fold compared to B/HPIV3 (Fig 1D).

Next, we compared the expression of the SARS-CoV-2 S by B/HPIV3/S-2P and B/HPIV3/S-6P by Western blotting. We infected Vero and A549 cells at a multiplicity of infection (MOI) of 0.1 plaque-forming unit (PFU) per cell with B/HPIV3 or the S-expressing versions, and prepared cell lysates 24 h after infection. Following SDS-PAGE (under reducing and denaturing conditions) and Western blotting using the hyperimmune antiserum against the secreted version of SARS-CoV-2 S-2P or purified HPIV3 virions, we detected the S protein in lysates at a size consistent with the uncleaved S0 precursor protein (Fig 1E and 1F).

We also evaluated the incorporation of the SARS-CoV-2 S-2P or S-6P protein into B/HPIV3 particles. Vero cell-grown viruses were purified by ultracentrifugation through sucrose gradients in three independent experiments. Western blotting revealed that both the S-2P and S-6P versions of the S protein were incorporated in B/HPIV3 particles (Fig 1G).

To evaluate the replication of B/HPIV3/S-2P and B/HPIV3/S-6P in the hamster model, groups of 28 hamsters were immunized intranasally with 5.0 log₁₀ PFU of B/HPIV3/S-2P, B/HPIV3/S-6P, or B/HPIV3 empty vector (Experiment #1, Fig 2A). On days 3, 5, and 7 post-immunization (pi), 5 hamsters per group were euthanized to evaluate vaccine virus replication in the respiratory tract. Nasal turbinates (NT) and lung tissue homogenates were prepared, and vaccine virus titers were determined by immunoplaque assay (Fig 2B–2C). Additional hamsters from each group were euthanized on days 3 (n = 1) and 5 (n = 2), and lungs and NT were analyzed by immunohistochemistry (IHC) (Fig 2D and 2E). Sera were obtained on day 26 pi from 10 animals per group to evaluate the antibody response.

Consistent with our previous results [17], B/HPIV3 replicated well in NT and lungs, with highest titers on day 3 pi (6.5 and 5.9 log₁₀ PFU/g, respectively) (Fig 2B, 2C). By day 5 pi, B/HPIV3 titers were 20-fold lower in NT, and about 10-fold lower in lungs. B/HPIV3/S-2P and B/HPIV3/S-6P also replicated well. On day 3 pi, titers of B/HPIV3/S-2P and B/HPIV3/S-6P in NT were 30 and 125-fold lower, respectively, compared to the B/HPIV3 empty vector; titers in lungs were 100- and 160-fold lower, respectively, than those of the B/HPIV3 empty vector. By day 5 pi, titers of B/HPIV3/S-2P and B/HPIV3/S-6P had increased in NT (25- and 50-fold, respectively) and lungs (6- and 20-fold, respectively), consistent with delayed replication in the upper and lower respiratory tract due to the presence of the insert. By day 7 pi, replication of B/HPIV3/S-2P, B/HPIV3/S-6P, and empty vector was very low or, in some animals, undetectable in NT or lungs, reflecting self-limiting replication of the vaccine candidates. We also determined the stability of S-2P and S-6P protein expression during in vivo replication by dual-staining immunoplaque assay; 96% of B/HPIV3/S-2P and 98% of B/HPIV3/S-6P plaques from NT and 95% of B/HPIV3/S-2P and 99% of B/HPIV3/S-6P plaques from lungs obtained on day 3 after immunization were positive for S protein; 90% and 98% (NT) or 87% and 96% (lungs) of B/HPIV3/S-2P and B/HPIV3/S-6P plaques from specimens obtained on day 5 after immunization were positive for the S protein, showing that expression of both versions of the S protein was stably maintained in vivo.

We also evaluated antigen expression in the NT and lungs by IHC. Representative IHC images from NT and lungs obtained on day 5 pi from one of two animals per group are shown in Fig 2D and 2E. B/HPIV3 antigen was present in columnar epithelial cells lining the nasal mucosa and the bronchioles in tissues from B/HPIV3, B/HPIV3/S-2P, and B/HPIV3/S-6P-immunized animals (red arrowheads, Fig 2D and 2E). SARS-CoV-2 S antigen in animals immunized with B/HPIV3/S-2P and B/HPIV3/S-6P similarly was present in columnar airway epithelial cells of NT and bronchioles (blue arrowheads, Fig 2D and 2E). The B/HPIV3 and SARS-CoV-2 S immunostaining patterns and intensities in tissues of B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals were generally similar. Thus, B/HPIV3/S-2P and B/HPIV3/S-6P efficiently infected and expressed the SARS-CoV-2 S protein in columnar epithelial cells of the nasal and bronchial mucosa.

To evaluate immunogenicity and protection against SARS-CoV-2 challenge, 45 hamsters per group were immunized intranasally with 5.0 log₁₀ PFU of B/HPIV3/S-2P, B/HPIV3/S-6P, or B/HPIV3 empty vector (Experiment #2, Fig 3A). We evaluated serum antibody responses 26 or 27 days after intranasal immunization. First, we measured SARS-CoV-2 S-specific serum IgG (Fig 3B) and IgA (Fig 3C) by ELISA using as antigens the secreted form of the S-2P protein (left panels) and a fragment (aa 319–591) of the S protein bearing the receptor-binding domain (RBD) (right panels). Intranasal immunization with B/HPIV3/S-2P and B/HPIV3/S-6P elicited strong serum anti-S and anti-RBD IgG and IgA responses in hamsters; the geometric mean (GMT) anti-S and anti-RBD IgG titers in the B/HPIV3/S-6P group were significantly higher (1.5-fold, P<0.01, and 2.4-fold, P<0.001) than those induced by B/HPIV3/S-2P. The GMT anti-RBD IgA titers in the B/HPIV3/S-6P group were also significantly higher than those induced by B/HPIV3/S-2P (1.4-fold, P<0.05), while the difference in anti-S IgA titers did not reach the level of significance. We also evaluated the serum antibody response in animals from Experiment #1 (n = 10 per group, S1A Fig), confirming that B/HPIV3/S-6P elicited higher anti-RBD IgG and IgA responses (S1B and S1C Fig; P<0.05). In Experiment #1, the differences in the anti-S IgG and IgA responses between B/HPIV3/S-2P and B/HPIV3/S-6P did not reach the level of statistical significance (S1B and S1C Fig).

Next, we evaluated the serum SARS-CoV-2-neutralizing antibody responses using the vaccine-matched SARS-CoV-2 strain WA1/2020 (S1D Fig for Experiment #1 and Fig 3D, left panel for Experiment #2). We also evaluated virus neutralizing antibody titers to two variants of concern (VoCs; B.1.1.7/Alpha and B.1.351/Beta) in live-virus neutralization assays (Fig 3D, right panels). B/HPIV3/S-2P and B/HPIV3/S-6P induced robust serum neutralizing antibody titers against all three viruses (Figs S1D and 3D). In addition, we randomly selected 10 sera per group to measure antibodies to B.1.617.2/Delta and B.1.1.529/Omicron variants in a pseudotype neutralization assay (in Biosafety level (BSL) 2); in both groups, we detected robust B.1.617.2/Delta pseudotype-inhibiting titers; B.1.1.529/Omicron pseudotype-inhibiting titers were about 10-fold lower than those detected in the B.1.617.2/Delta pseudotype assay. As expected, SARS-CoV-2-neutralizing antibodies were not detected in animals immunized with the B/HPIV3 empty vector. We also evaluated the serum neutralizing antibody titers to HPIV3 in animals from Experiment #2 (Fig 3F) and #1 (S1E Fig), showing a strong neutralizing immune response in all groups. The B/HPIV3 empty-vector immunized animals exhibited the strongest anti-vector immune response (Experiment #2: 1.6 and 2.1-fold higher than those induced by B/HPIV3/S-2P and B/HPIV3/S-6P, P<0.0001).

Finally, we evaluated the ability of serum antibodies from the B/HPIV3/S-2P- and B/HPIV3/S-6P-immunized hamsters to block binding of soluble angiotensin-converting enzyme 2 (ACE2) to SARS-CoV-2 S proteins of several lineages. This sensitive binding inhibition assay is based on electrochemiluminescence and serves as a high-throughput alternative to BSL3 SARS-CoV-2 neutralization assays. At a serum dilution of 1:20, antibodies from B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals completely inhibited ACE2 binding to recombinant S proteins derived from the vaccine-matched Wuhan-Hu-1 or VoCs B.1.1.7/Alpha and B.1.617.2/Delta (Fig 3G), and they also efficiently inhibited ACE2 binding to S proteins of B.1.351/Beta (Fig 3G) or B.1.640.2 (S2 Fig) (93.6 to 97.7 median % inhibition). Furthermore, serum antibodies induced by B/HPIV3/S-2P and B/HPIV3/S-6P also inhibited ACE2 binding to S proteins of the B.1.1.529/Omicron BA.1, BA.2, BA.2.12.1, BA.3, BA.4 and BA.5 sublineages, even though the rate of inhibition was lower (54.1 to 91.1 median % inhibition, Figs 3G and S2). No ACE2 binding inhibition was detectable in sera from the B/HPIV3 empty vector-immunized hamsters. Thus, intranasal immunization with B/HPIV3/S-2P and B/HPIV3/S-6P induces a potent neutralizing antibody response with broad activity against SARS-CoV-2 variants, as well as to HPIV3.

Hamsters from each of the 3 groups in Experiment #2 (n = 45 per group, immunized with B/HPIV3, B/HPIV3/S-2P, or B/HPIV3/S-6P, respectively) were divided randomly into 3 subgroups (n = 15 per subgroup). On day 30 after intranasal immunization with B/HPIV3, B/HPIV3/S-2P, or B/HPIV3/S-6P, each of the 3 subgroups were challenged intranasally with 4.5 log₁₀ 50% tissue culture infectious doses (TCID₅₀) of the vaccine-matched SARS-CoV-2 strain WA1/2020 or VoCs B.1.1.7/Alpha or B.1.351/Beta (n = 15 per challenge virus subgroup, Fig 3A). After challenge with WA1/2020, B.1.1.7/Alpha or B.1.351/Beta, animals previously immunized with B/HPIV3 empty-vector control gradually lost weight over a period of about 7 days (Fig 4A–4C, green lines), and generally started recovering weight beginning on day 7 or 8. Following challenge with WA1/2020, one animal immunized with B/HPIV3 empty-vector control reached the humane study endpoint of 25% weight loss on day 11. On the other hand, hamsters previously immunized with B/HPIV3/S-2P (orange lines) or B/HPIV3/S-6P (blue lines) continued gaining weight after challenge with WA1/2020 and B.1.1.7/Alpha (Fig 4A and 4B). Following challenge with B.1.351/Beta (Fig 4C), the weight gain in B/HPIV3/S-2P or B/HPIV3/S-6P-immunized animals was somewhat less uniform between days 2 and 7 after challenge, but the mean body weight in both vaccine groups was significantly higher compared to the corresponding B/HPIV3 empty-vector control-immunized group over this period.

To further evaluate protection from challenge with SARS-CoV-2 WA1/2020, B.1.1.7/Alpha or B.1.351/Beta, we euthanized 5 animals per group on days 3 and 5 after challenge, and extracted RNA from lung homogenates to evaluate the relative expression of inflammatory cytokines C-X-C motif ligand 10 (CXCL10), myxovirus resistance protein 2 (Mx2), or interferon lambda (IFN-L) in response to SARS-CoV-2 challenge by TaqMan assay (Fig 4D–4F). In previous studies, we had identified CXCL10 and Mx2 as the most strongly induced interferon response genes in hamsters following intranasal inoculation with SARS-CoV-2 [17], serving as biomarkers, together with IFN-L, for the SARS-CoV-2 induced cytokine storm and providing correlates of disease severity [24]. Indeed, after challenge with WA1/2020, B.1.1.7/Alpha or B.1.351/Beta, we detected strong expression of CXCL10, Mx2, or IFN-L in all B/HPIV3 empty-vector control immunized animals (Fig 4D–4F). However, after challenge of the B/HPIV3/S-2P or B/HPIV3/S-6P-immunized animals with any of these three challenge strains, we found only low or baseline expression of CXCL10, Mx2, or IFN-L, confirming that intranasal immunization with B/HPIV3/S-2P or B/HPIV3/S-6P was protective against cytokine or IFN induction after challenge with the vaccine-matched WA1/2020 strain, or representatives of B.1.1.7/Alpha or B.1.351/Beta variants.

We also determined SARS-CoV-2 challenge virus titers after challenge with strain WA1/2020 or VoCs (B.1.1.7/Alpha or B.1.351/Beta) in tissue homogenates of NT and lung tissues obtained on days 3 and 5 from 5 animals per group (Fig 5A–5C). After challenge of B/HPIV3 empty-vector immunized animals with WA1/2020, B.1.1.7/Alpha or B.1.351/Beta variants, we detected high SARS-CoV-2 titers in lungs on both days, with peak GMTs of 8.0 log₁₀ TCID₅₀ per g (WA1/2020, Fig 5A, left panel) or 7.4 log₁₀TCID₅₀/g (B.1.1.7/Alpha and B.1.351/Beta; Fig 5B and 5C, left panels). Peak GMTs in NT reached 5.6, 5.8, or 4.9 log₁₀TCID₅₀/g for WA1/2020, B.1.1.7/Alpha or B.1.351/Beta on day 3 (Fig 5A–5C, right panels). In contrast, B/HPIV3/S-2P and B/HIPV3/S-6P-immunized hamsters did not have any WA1/2020, B.1.1.7/Alpha or B.1.351/Beta challenge virus detectable in lungs and NT on days 3 and 5 after challenge, with the exception of 5 animals with detectable virus on day 3 after challenge only; specifically, two B/HPIV3/S-2P-immunized hamsters had WA1/2020 challenge virus detectable in NT, and one hamster immunized with B/HPIV3/S-2P and two hamsters immunized with B/HPIV3/S-6P had B.1.351/Beta challenge virus detectable in both NT and lungs on day 3.

Next, we determined the SARS-CoV-2 viral loads in lungs after challenge. RNA from lung homogenates was evaluated by TaqMan assays to detect genomic N (gN) and E (gE) SARS-CoV-2 RNA, as well as subgenomic E (sgE) RNA. sgE RNA served as an indicator for active replication of challenge virus (Fig 5D–5F). In tissues obtained from all B/HPIV3 empty-vector immunized animals on days 3 or 5 after SARS-CoV-2 challenge, we detected very high viral loads (ranging between 9 and 11 log₁₀ copies per g) of gN, gE, and sgE per g of lung tissue. The high levels of sgE copy numbers confirmed high levels of WA1/2020, B.1.1.7/Alpha or B.1.351/Beta challenge virus replication in the empty-vector immunized group. In the B/HPIV3/S-2P and B/HPIV3/S-6P-immunized groups, the levels of genomic RNA were significantly lower on days 3 and 5 after challenge with WA1/2020, B.1.1.7/Alpha or B.1.351/Beta. The levels of subgenomic RNA were also lower or undetectable in the B/HPIV3/S-2P and B/HPIV3/S-6P-immunized groups. Specifically, following WA1/2020 challenge, low levels of sgE RNA were only detectable in one B/HPIV3/S-2P and two B/HPIV3/S-6P-immunized animals on day 3 after challenge; after B.1.351/Beta challenge, low levels of subgenomic RNA were detectable in 2 animals per group on day 3. After B.1.1.7/Alpha challenge of B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals, no sgE RNA was detectable on day 3; by day 5 after challenge, no subgenomic RNA was detectable in any of the SARS-CoV-2 challenge subgroups of B/HPIV3/S-2P or B/HPIV3/S-6P-immunized animals. Thus, intranasal immunization with B/HPIV3/S-2P or B/HPIV3/S-6P was protective in hamsters against challenge virus replication of a vaccine-matched strain of SARS-CoV-2, and two VoCs (Alpha and Beta).

On day 21 after SARS-CoV-2 challenge, we collected serum from the remaining animals (n = 5 per challenge group) and determined the neutralizing serum antibody titers against WA1/2020, B.1.1.7/Alpha or B.1.351/Beta in live-virus neutralization assays performed at BSL3 (Fig 6A). We found that (i) challenge with WA1/2020, B.1.1.7/Alpha or B.1.351/Beta boosted the WA1/2020 and B.1.1.7/Alpha neutralizing antibody titers of B/HPIV3/S-2P- and B/HPIV3/S-6P-immunized animals (Figs 3D and 6A, left and middle panel). (ii) The strongest boost was detected in B/HPIV3/S-6P-immunized, WA1/2020 challenged animals that exhibited three-fold higher anamnestic serum neutralizing antibody titers against WA1/2020 than B/HPIV3/S-2P-immunized animals (P<0.05, Fig 6A left panel, red label). (iii) WA1/2020 and B.1.1.7/Alpha challenge also boosted the B.1.351/Beta cross-neutralizing titers of the B/HPIV3/S-2P- and the B/HPIV3/S-6P-immunized animals (Figs 3D and 6A, right panel); (iv) surprisingly, challenge with the vaccine-matched WA1/2020 virus or a more closely antigenically related virus (B.1.1.7/Alpha) boosted the neutralizing activity to a more antigenically distinct virus, B.1.351/Beta, more effectively than challenge with the antigenically distant B.1.351/Beta virus itself (Fig 6A, right panel).

We also evaluated the ability of the sera from the challenged animals to block binding of a soluble version of ACE2 to S proteins derived from various SARS-CoV-2 lineages, using the ACE2 binding inhibition assay as a high-throughput alternative to BSL3 virus neutralization assays. Twenty-one days after challenge with WA1/2020, B.1.1.7 or B.1.351, sera from B/HPIV3 empty-vector control immunized/SARS-CoV-2 challenged animals inhibited ACE2 binding of S proteins from all 15 lineages and sublineages tested, showing that in the empty-vector control animals, each of the three challenge viruses induced primary antibody responses with broad ACE2 binding inhibition activity to S proteins. Specifically, at the dilution tested, post-challenge sera from B/HPIV3 empty-vector immunized animals completely blocked ACE2 binding to S proteins of the challenge virus-matched lineages (Fig 6B, top panels, red labels), and to heterologous WA1/2020 and B.1.1.7/Alpha-derived S proteins (Figs 6B and S3, black labels); ACE2 binding inhibition by these sera to the B.1.617.2/Delta-derived S protein also was nearly complete, while the median ACE2 binding inhibition of post-challenge sera from B/HPIV3 control-immunized animals to S proteins of Omicron variants ranged between 45 and 90% (Figs 6B and S3).

Sera from B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals further challenged with WA1/2020, B.1.1.7 or B.1.351 also completely inhibited ACE2 binding to S proteins of WA1/2020, B.1.1.7/Alpha, B.1.351/Beta, B.1.617.2/Delta lineages (Fig 6B, top row panels), or of the B.1.640.2 lineage (S3 Fig, bottom, right panel). In ACE2 binding inhibition assays to S proteins of all 10 Omicron sublineages tested, WA1/2020, B.1.1.7/Alpha and B.1.351/Beta post-challenge sera of B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals showed significantly stronger binding inhibition compared to post-challenge sera from B/HPIV3 empty-vector control immunized animals (median ranging between 82 and 99%, Figs 6B and S3), showing that intranasal immunization with B/HPIV3/S-2P and B/HPIV3/S-6P was effective in priming for strong secondary antibody responses with ACE2 binding inhibition activity. Thus, homologous or heterologous SARS-CoV-2 challenge increased the magnitude and breadth of the anamnestic serum antibody responses in B/HPIV3/S-2P and B/HPIV3/S-6P-immunized animals.

All animal studies were approved by the NIAID Animal Care and Use Committee.

Human lung epithelial A549 cells (ATCC CCL-185) were grown in F-12K medium (ATCC) with 5% fetal bovine serum (FBS). LLC-MK2 rhesus monkey kidney cells (ATCC CCL-7), African green monkey kidney Vero cells (ATCC CCL-81), or Vero E6 cells (ATCC CRL-1586) were grown in Opti-MEM (Thermo Fisher) with 5% FBS. Vero E6 cells were used for SARS-CoV-2 neutralization assays, and to titrate the SARS-CoV-2 challenge viruses.

The SARS-CoV-2 USA-WA1/2020 challenge virus stock was grown on Vero E6 cells stably expressing human TMPRSS2 [17]. The USA/CA_CDC_5574/2020 isolate (lineage B.1.1.7/Alpha variant, GISAID: EPI_ISL_751801; sequence deposited by CDC, isolate obtained from CDC) and the USA/MD-HP01542/2021 isolate (lineage B.1.351/Beta variant, GISAID: EPI_ISL_890360; sequence deposited by Morris et al., Johns Hopkins University (JHU), isolate obtained from Andrew Pekosz, JHU, Baltimore, MD) were also passaged on Vero E6 cells stably expressing TMPRSS2 [17]. Titration of SARS-CoV-2 was performed by determination of the TCID₅₀ in Vero E6 cells [33]. All experiments with SARS-CoV-2 were conducted in Biosafety Level (BSL)-3 containment laboratories approved for use by the US Department of Agriculture and CDC.

Virus stocks of recombinant B/HPIV3 [8,9] and B/HPIV3 vectors were propagated on Vero cells at 32°C and titrated by dual-staining immunoplaque assay [17].

Briefly, Vero cells in 24-well plates were inoculated with 10-fold serially diluted samples. Two hours later, monolayers were overlaid with culture medium containing 0.8% methylcellulose and incubated at 32°C. Six days later, monolayers were fixed with 80% methanol, and immunostained with an HPIV3-specific rabbit hyperimmune serum to detect B/HPIV3 antigens, and a goat hyperimmune serum against the secreted version of SARS-CoV-2 S-2P to detect co-expression of the S protein. An infrared-dye conjugated donkey anti-rabbit IRDye680 IgG and a donkey anti-goat IRDye800 IgG were used as secondary antibodies. Stained plates were scanned with the Odyssey infrared imaging system (LiCor). Fluorescent staining for PIV3 proteins and SARS-CoV-2 S was visualized in green and red, respectively, generating yellow plaque staining when merged [17].

Sub-confluent Vero and A459 cells in 6-well plates were infected in triplicate with indicated viruses at a multiplicity of infection (MOI) of 0.01 PFU per cell. After adsorption for 2 h, cells were washed to remove the virus inoculum, and 3 ml of fresh media were added. Cells were incubated at 32°C for 7 days. Every 24 h, 0.5 ml aliquots of culture medium were collected and snap-frozen in dry ice, and 0.5 ml of fresh medium was added to each well. Virus aliquots from each time point were titrated side-by-side on 24-well plates of Vero cells as described above.

Six-well plates of Vero or A549 cells were inoculated using an MOI of 1 PFU per cell with B/HPIV3, B/HPIV3/S-2P, and B/HPIV3/S-6P and incubated at 32°C. At 48 h post-inoculation, cells were washed once with cold PBS and lysed using 300 μl LDS lysis buffer (Thermo Fisher Scientific) containing NuPAGE reducing reagent (Thermo Fisher Scientific). Cell lysates were further passed through QIAshredder columns (Qiagen), heated for 10 min at 95°C and separated on 4–12% Bis-Tris NuPAGE gels (Thermo Fisher Scientific) in the presence of antioxidant (Thermo Fisher Scientific). The resolved proteins were transferred to polyvinylidene difluoride membranes which were blocked with blocking buffer (LiCor) for 1 h at room temperature. Then, blocked membranes were incubated with a goat hyperimmune serum to SARS-CoV-2 S and a rabbit polyclonal hyperimmune sera against purified HPIV3 in blocking buffer overnight at 4°C. A mouse monoclonal antibody to GAPDH (Sigma) was included as a loading control. Membranes were incubated with infrared dye-labeled secondary antibodies (donkey anti-rabbit IgG IRDye 680, donkey anti-goat IgG IRDye 800 or IRDye 680, and donkey anti-mouse IgG IRDye 800, LiCor). Intensities of individual protein bands from the acquired images were quantified after background correction using Image Studio software (LiCor).

The protein composition of virus particles was also analyzed. To do so, viruses were grown on Vero cells and the supernatant was purified by centrifugation through 30%/60% sucrose gradients. The sucrose-purified virus preparations were gently pelleted by centrifugation to remove sucrose as described previously [34]. The protein concentration of the purified virus preparations was determined using a BCA protein assay kit (Thermo Fisher Scientific) prior to the addition of lysis buffer. One μg of protein per lane was used for SDS-PAGE and Western blotting.

In Experiment #1, groups (n = 28) of 5- to 6-week-old male golden Syrian hamsters (Envigo Laboratories, Frederick, MD) were anesthetized and inoculated intranasally with 100 μl of Leibovitz’s L-15 medium (Thermo Fisher Scientific) containing 5 log₁₀ PFU of B/HPIV3, B/HPIV3/S-2P, or B/HPIV3/S-6P viruses. On days 3, 5, and 7 post-inoculation, 5 hamsters per group were euthanized by CO₂ inhalation, and nasal turbinates and lungs were collected to evaluate virus replication. NT and lung tissue samples for histology were obtained from one and two additional hamsters per group on days 3 and 5, respectively. For quantification of virus replication, tissues were homogenized in Leibovitz’s L-15 medium, and clarified homogenates were analyzed by dual-staining immunoplaque assay on Vero cells as described above. On day 26 post-immunization, sera were collected from the remaining 10 animals per group to evaluate the immunogenicity of the vaccine candidates. HPIV3 vector-specific neutralizing antibodies were detected by a 60% plaque reduction neutralization test (PRNT₆₀) [35] on Vero cells in 24-well plates using HPIV3 expressing the eGFP protein. To determine the serum neutralizing antibody response to SARS-CoV-2, two-fold dilutions of heat-inactivated hamster sera were tested in a microneutralization assay for the presence of antibodies that neutralized the replication of 100 TCID₅₀ of SARS-CoV-2 in Vero E6 cells as previously described [17]. Serum IgG antibodies to a secreted form of the S-2P protein [23] or its RBD [36] were measured by ELISA [17], and serum IgA antibodies to the secreted form of S-2P or RBD were measured by europium ion-enhanced DELFIA-TRF assay (Perkin Elmer) following the supplier’s protocol.

In Experiment #2, groups (n = 45) of 5- to 6-week-old male golden Syrian hamsters were immunized intranasally with 5 log₁₀ PFU of B/HPIV3, B/HPIV3/S-2P, or B/HPIV3/S-6P as described above. Sera were obtained on days 26 or 27. Hamsters in each immunization group were randomly distributed to 3 challenge virus subgroups of 15 animals per subgroup, and challenged intranasally with 4.5 log₁₀ TCID₅₀ of SARS-CoV-2 WA1/2020, the USA/CA_CDC_5574/2020 isolate (lineage B.1.1.7/Alpha variant) or USA/MD-HP01542/2021 (lineage B.1.351/Beta variant) in 100 μl. Five hamsters from each challenge virus subgroup were euthanized by CO₂ inhalation on days 3 and 5 after challenge, and tissues were collected to evaluate challenge virus replication. The presence of challenge virus in clarified tissue homogenates was evaluated later by TCID₅₀ assay on Vero E6 cells. The remaining 5 hamsters from each challenge virus subgroup were euthanized on day 21 after challenge, and sera were collected for antibody analysis.

Lung tissue samples from hamsters were fixed in 10% neutral buffered formalin, processed through a Leica ASP6025 tissue processor (Leica Biosystems), and embedded in paraffin. Five μm tissue sections were stained with hematoxylin and eosin for routine histopathology. For IHC evaluation, sections were deparaffinized and rehydrated. After epitope retrieval, sections were labeled with a goat hyperimmune serum to anti-SARS-CoV-2 S (N25-154) at 1:1000, and a rabbit polyclonal anti-HPIV3 serum [37] at 1:500. Chromogenic staining was carried out on the Bond RX platform (Leica Biosystems) according to manufacturer-supplied protocols. Detection with DAB chromogen was completed using the Bond Polymer Refine Detection kit (Leica Biosystems). The VisUCyte anti-goat horseradish peroxidase (HRP) polymer (R&D Systems) replaced the standard Leica anti-rabbit HRP polymer from the kit to bind the goat anti-SARS-CoV-2 S antibodies. Slides were finally cleared through gradient alcohol and xylene washes prior to mounting. Sections were examined by a board-certified veterinary pathologist using an Olympus BX51 light microscope. Images were acquired using an Olympus DP73 camera.

Single-round luciferase-expressing pseudoviruses were generated by co-transfecting plasmids encoding SARS-CoV-2 S of isolate B.1.617.2/Delta or B.1.1.529/Omicron, luciferase reporter (pHR’ CMV Luc), lentivirus backbone (pCMV ΔR8.2), and human transmembrane protease serine 2 (TMPRSS2) at a ratio of 1:20:20:0.3 into HEK293T/17 cells (ATCC CRL-11268) using transfection reagent LiFect293. Three days post-transfection, supernatants were collected and centrifugated at 478 x g for 10 min to remove cell debris and filtered through a 0.45 μm filter. Pseudoviruses were then aliquoted and titrated. For the antibody neutralization assay, 6-point, 5-fold dilution series were prepared in DMEM medium supplemented with 10% FBS, 1% Pen/Strep and 3 μg/ml puromycin. Fifty μl of the diluted sera were mixed with 50 μl of diluted pseudoviruses in 96-well plates and incubated for 30 min at 37°C. Then, ten thousand of ACE2-expressing 293T-cells (293T-hACE2.MF stable cell line) per well were added. Three days later, cells were lysed with Bright-Glo luciferase assay substrate (Promega), and luciferase activity (relative light units, RLU) was evaluated. The percent neutralization was normalized relative to uninfected cells as 100% neutralization and cells infected with only pseudoviruses as 0% neutralization. Fifty percent inhibitory concentration (IC₅₀) titers were determined using a log (agonist) vs. normalized response (variable slope) nonlinear function in Prism v9 (GraphPad).

A 10-plex ACE2 binding inhibition assay was performed as an alternative to a BSL3 SARS-CoV-2 neutralization assay [SARS-CoV-2 Panel 25 and 27 (K15586U and K15609U, respectively), Meso Scale Diagnostics (MSD)], following the manufacturer’s instructions. The panels contain 96-well, 10-spot plates coated with soluble ectodomains of spike proteins from the vaccine-matched SARS-CoV-2 (Wuhan-Hu-1, identical amino acid sequence as WA1/2020) and VoCs (B.1.1.7/Alpha, B.1.351/Beta, B.1.617.2/Delta, B.1.1.529/Omicron BA.1, BA.1+S[R346K], BA.1+S[L452R], BA.2, BA.2+S[L452M], BA.2+S[L452R], BA.2.12.1, BA.3, BA.4, BA.5 sublineages) and a variant under monitoring (B.1.640.2). In brief, plates were blocked with MSD blocker A for 30 min and washed with MSD wash buffer. Heat-inactivated hamster sera were diluted 1:20 in diluent and added to duplicate wells. After a one-hour incubation, Sulfo-tag labelled ACE2 was added to each well. All incubations were performed on a plate shaker at room temperature. Following a one-hour incubation, plates were washed, MSD GOLD electrochemiluminescence read buffer B was added, and chemiluminescence of bound Sulfo-tag-ACE2 was detected using a MESO QuickPlex SQ 120 reader (MSD). The average electrochemiluminescence signals in duplicate wells for each serum were determined, as well as the maximum electrochemiluminescence signals of ACE2/S protein binding in no-serum control wells. The ACE2 neutralizing activity of each serum is represented as percent inhibition relative to no-serum controls.

Total RNA was extracted from 0.1 ml of lung homogenates (0.1 g/ml) using the TRIzol LS Reagent and Phasemaker Tubes Complete System (Thermo Fisher) along with the PureLink RNA Mini Kit (Thermo Fisher) following the manufacturer’s instructions, and eluted in 0.1 ml RNase-free water. Total RNA was also extracted from lung homogenates of five control hamsters (non-immunized and non-challenged) in the same manner. cDNA was synthesized from 7 μl of RNA using the High-Capacity RNA-to-cDNA Kit (Thermo Fisher). TaqMan assays (Thermo Fisher) specific for hamster (Mesocricetus auratus) CXCL10, Mx2, or IFN- L [38,39] were performed using the TaqMan Fast Advanced Master Mix (Thermo Fisher), and hamster beta-actin was included as a housekeeping gene. qPCR results were analyzed using the comparative threshold cycle (ΔΔC_T) method, normalized to beta-actin, and expressed as fold changes over the average expression of five uninfected, unchallenged hamsters.

To detect viral genomic N (gN), E (gE) RNA, and subgenomic E (sgE) mRNA of SARS-CoV-2 challenge virus WA1/2020, B.1.1.7/Alpha, or B.1.351/Beta in the lung homogenates, cDNA was synthesized from 8 μl of total RNA extracted as described above and TaqMan qPCRs for genomic N, E, and sgE were performed using previously-described primers and probes [40–42] and the TaqMan RNA-to-Ct 1-Step Kit (Thermo Fisher). Assays were performed on a QuantStudio 7 Pro real-time PCR system (Thermo Fisher). Standard curves were generated using serially diluted pcDNA3.1 plasmids containing gN, gE, or sgE sequences. The sensitivity of the TaqMan assay was 200 copies, corresponding to a limit of detection of 5.4 log₁₀ copies per g of tissue.

Data sets were assessed for significance using one-way ANOVA with Tukey’s or Sidak’s multiple comparisons test, two-way ANOVA with Tukey’s post-hoc test or Mixed-effects analysis with Tukey’s post-hoc test using Prism 9 (GraphPad Software). Data were only considered significant at P ≤ 0.05. Details on the statistical comparisons can be found in the figures, figure legends, and results. Values of n are provided in the text and in the figure legends.

The numerical data used in all figures are included in S1 Data.